A metabolite glue predicts inverse coupling of AICAR to one-carbon supply

Abstract

Formate, an output of mitochondrial one-carbon metabolism, drives a switch from low to high adenine nucleotide levels in proliferating cells. The kinetic model of Oizel et al 2020 reproduced this switch under the assumption that de novo purine synthesis is slaved to biosynthetic demand. Two independent studies have since identified the physical controller that this assumption stands in for: the ADP-ribose pyrophosphatase NUDT5 binds and inhibits phosphoribosyl pyrophosphate amidotransferase (PPAT), the rate-limiting enzyme of purine synthesis, with AMP acting as an allosteric metabolite glue and the substrate PRPP competing to dissociate the complex. We replace the demand-slaving assumption with an explicit PRPP pool and a glue-gated PPAT step, calibrated entirely against the published binding data. The model makes a sharp, falsifiable prediction: the purine precursor AICAR is coupled inversely to one-carbon availability, rising as one-carbon units become scarce and clearing when they are abundant. This inverse coupling is a consequence of the glue: with the feedback removed -- the genetic state of a NUDT5 knockout -- the model instead predicts AICAR rising with one-carbon availability, opposite in sign. The prediction reconciles the AICAR measurements of Oizel eat al 2020 without having been fitted to them, and the model independently reproduces the fall of the PRPP pool reported in NUDT5-deleted cells. We propose experiments that discriminate the glue from a constitutive throttle, and discuss the consequences for AMPK signalling and for the cytotoxicity of thiopurine and antifolate chemotherapeutics, whose action converges on this same PPAT--NUDT5 node.